Anker’s SleepLab Pro Uses Radar to Track Rest and Automate Your Bedroom

For anyone whose smart home runs on presence data, this is a new sensor category creeping into a corner people already overload with cameras and wearables. The same IFA floor also produced Anker's MindBase, an on-device AI hub running Matter 1.5 with enough radios to actually unify the mesh — and that's the piece automation architects should be staring at.

SleepLab Pro: a presence sensor wearing a speaker costume

The pitch is simple: drop it on the nightstand, let it pulse 60GHz radar at your chest cavity, and it pulls respiratory rate, heart rate, and sleep stages without strapping anything to your body. It also plays the usual sleep-audio palette and runs a gradually-shifting light, and Anker claims it logs which combination of those settings actually put you under so you can replay them on demand. No subscription required, which already puts it ahead of most mattress pads and rings in the automation ledger.

From an automation standpoint, the interesting payload isn't the sleep data — it's presence. A bedside sensor that knows you're in the bed, and per Anker filters out the partner, is the missing trigger for half the routines people currently brute-force with phone GPS or wearable states. Think: HVAC pre-cool only when actually under the covers, lighting scenes that branch on sleep stage rather than time-of-day, white noise that backs off the moment REM hits. The radar doesn't share that data with HomeKit or anything else until the device ships, though — battery or wired is unconfirmed, the Bluetooth driver specs haven't been published, and the bedmate-detection claim is hand-waved in press materials.

File this under wait-but:

• No MSRP, no release window finer than later this year.
• No published APIs, Matter endpoints, or local integration hooks yet — anything beyond the vendor app is wishful thinking right now.
• 60GHz sleep staging on a non-worn device is a measurement claim that hasn't been independently validated against polysomnography in any source available.

MindBase: the part that actually matters for the mesh

The protocol story lives in the broader IFA bundle Anker dropped alongside the speaker. The MindBase is pitched as a "central brain" hub, but the spec sheet reads like a checklist automation architects have been hand-rolling for years: Matter 1.5 controller, dual-band Wi-Fi, Bluetooth, Thread, Zigbee, Apple HomeKit Secure Video, Alexa, Google Home, and Home Assistant. Up to 48TB of expandable local storage, on-device LLM for event processing, sixteen wireless cameras plus four wired, and a new Smart Security Shield sensor using 180-degree dual radar with phone-stored digital credentials to recognize household members and squash notification spam.

In principle, that's the kind of hub that lets you rip out the cloud round-trips for camera AI and run detection logic on the LAN. Anker's prior encryption scandal is the obvious asterisk — local-only and end-to-end are claims worth pressure-testing before any third-party camera gets pointed at the front door. The first horse in the gate is the TrackLight Cam S1, an auto-tracking 4K camera with 180-degree dual radar that ships as part of an Anker MindBase Security Kit. The wired Eufy S4 video doorbell lands this week at $299.99 with 180-by-180 panoramic capture, AI tracking, 3K recording, 64GB onboard, Wi-Fi 6, and radar-plus-PIR motion sensing. A $70 window camera arrives later in the year.

What to wire up first when these actually ship

Skip the SleepLab Pro for routine-building purposes until someone publishes MQTT, Matter, or local API behavior. Until then it lives in its own walled app, which defeats the purpose for anyone running Home Assistant. The MindBase is the more interesting acquisition if the spec survives independent teardown — a Matter 1.5 controller with Thread and Zigbee on the same box, plus local LLM inference for camera events, is the closest thing the Eufy line has offered to a real local-first hub. Hold the routine that pipes camera events into Node-RED or n8n until at least one user-level teardown confirms on-device processing is genuinely on-device and not a cloud relay dressed up in different clothes. Everything else — Matter device category coverage, the sensor ceiling of 34 units, the family-recognition credential flow — reads like the architecture you'd build yourself if you had the time and the BOM budget.